Mud­dling Along Is More Likely Than Dystopia

Jeffrey HeningerOct 20, 2023, 9:25 PM
88 points
10 comments8 min readLW link

What’s Hard About The Shut­down Problem

johnswentworthOct 20, 2023, 9:13 PM
98 points
33 comments4 min readLW link

Holly El­more and Rob Miles di­alogue on AI Safety Advocacy

Oct 20, 2023, 9:04 PM
162 points
30 comments27 min readLW link

TOMORROW: the largest AI Safety protest ever!

Holly_ElmoreOct 20, 2023, 6:15 PM
105 points
26 comments2 min readLW link

The Overkill Con­spir­acy Hypothesis

ymeskhoutOct 20, 2023, 4:51 PM
26 points
8 comments7 min readLW link

I Would Have Solved Align­ment, But I Was Wor­ried That Would Ad­vance Timelines

307thOct 20, 2023, 4:37 PM
122 points
33 comments9 min readLW link

In­ter­nal Tar­get In­for­ma­tion for AI Oversight

Paul CologneseOct 20, 2023, 2:53 PM
15 points
0 comments5 min readLW link

On the proper date for sols­tice celebrations

jchanOct 20, 2023, 1:55 PM
16 points
0 comments4 min readLW link

Are (at least some) Large Lan­guage Models Holo­graphic Me­mory Stores?

Bill BenzonOct 20, 2023, 1:07 PM
11 points
4 comments6 min readLW link

Mechanis­tic in­ter­pretabil­ity of LLM anal­ogy-making

SergiiOct 20, 2023, 12:53 PM
2 points
0 comments4 min readLW link
(grgv.xyz)

How To So­cial­ize With Psy­cho(lo­gist)s

SableOct 20, 2023, 11:33 AM
37 points
11 comments3 min readLW link
(affablyevil.substack.com)

Re­veal­ing In­ten­tion­al­ity In Lan­guage Models Through AdaVAE Guided Sampling

jdpOct 20, 2023, 7:32 AM
119 points
15 comments22 min readLW link

Fea­tures and Ad­ver­saries in MemoryDT

Oct 20, 2023, 7:32 AM
31 points
6 comments25 min readLW link

AI Safety Hub Ser­bia Soft Launch

DusanDNesicOct 20, 2023, 7:11 AM
64 points
1 comment3 min readLW link
(forum.effectivealtruism.org)

An­nounc­ing new round of “Key Phenom­ena in AI Risk” Read­ing Group

Oct 20, 2023, 7:11 AM
15 points
2 comments1 min readLW link

Un­pack­ing the dy­nam­ics of AGI con­flict that sug­gest the ne­ces­sity of a premp­tive pivotal act

Eli TyreOct 20, 2023, 6:48 AM
63 points
2 comments8 min readLW link

Geno­cide isn’t Decolonization

robotelvisOct 20, 2023, 4:14 AM
33 points
19 comments5 min readLW link
(messyprogress.substack.com)

Try­ing to un­der­stand John Went­worth’s re­search agenda

Oct 20, 2023, 12:05 AM
93 points
13 comments12 min readLW link

Boost your pro­duc­tivity, hap­piness and health with this one weird trick

ajc586Oct 19, 2023, 11:30 PM
9 points
9 comments1 min readLW link

A Good Ex­pla­na­tion of Differ­en­tial Gears

Johannes C. MayerOct 19, 2023, 11:07 PM
48 points
4 comments1 min readLW link
(youtu.be)

Even­ing Wiki(pe­dia) Workout

mcintOct 19, 2023, 9:29 PM
1 point
1 comment1 min readLW link

New roles on my team: come build Open Phil’s tech­ni­cal AI safety pro­gram with me!

Ajeya CotraOct 19, 2023, 4:47 PM
83 points
6 comments4 min readLW link

[Question] In­finite tower of meta-probability

fryolysisOct 19, 2023, 4:44 PM
6 points
5 comments3 min readLW link

A NotKillEvery­oneIsm Ar­gu­ment for Ac­cel­er­at­ing Deep Learn­ing Research

Logan ZoellnerOct 19, 2023, 4:28 PM
−6 points
6 comments5 min readLW link
(midwitalignment.substack.com)

Knowl­edge Base 5: Busi­ness model

iwisOct 19, 2023, 4:06 PM
−4 points
2 comments1 min readLW link

AI #34: Chip­ping Away at Chip Exports

ZviOct 19, 2023, 3:00 PM
36 points
19 comments59 min readLW link
(thezvi.wordpress.com)

Is Yann LeCun straw­man­ning AI x-risks?

Chris_LeongOct 19, 2023, 11:35 AM
26 points
4 comments1 min readLW link

[Video] Too much Em­piri­cism kills you

Johannes C. MayerOct 19, 2023, 5:08 AM
19 points
0 comments1 min readLW link
(youtu.be)

Are hu­mans mis­al­igned with evolu­tion?

Oct 19, 2023, 3:14 AM
42 points
13 comments18 min readLW link

Brains, Planes, Blimps, and Algorithms

ai danOct 18, 2023, 9:26 PM
1 point
0 comments6 min readLW link

The (par­tial) fal­lacy of dumb superintelligence

Seth HerdOct 18, 2023, 9:25 PM
38 points
5 comments4 min readLW link

[Question] Does AI gov­er­nance needs a “Fed­er­al­ist pa­pers” de­bate?

azsantoskOct 18, 2023, 9:08 PM
40 points
4 comments1 min readLW link

Me­tac­u­lus Launches Con­di­tional Cup to Ex­plore Linked Forecasts

ChristianWilliamsOct 18, 2023, 8:41 PM
9 points
0 commentsLW link
(www.metaculus.com)

AI Safety 101 : Re­ward Misspecification

markovOct 18, 2023, 8:39 PM
32 points
4 comments31 min readLW link

2023 East Coast Ra­tion­al­ist Megameetup

ScrewtapeOct 18, 2023, 8:33 PM
8 points
0 comments1 min readLW link

Su­perfore­cast­ing the premises in “Is power-seek­ing AI an ex­is­ten­tial risk?”

Joe CarlsmithOct 18, 2023, 8:23 PM
31 points
3 comments5 min readLW link

The Real Fan­fic Is The Friends We Made Along The Way

EneaszOct 18, 2023, 7:21 PM
92 points
1 comment27 min readLW link1 review
(deathisbad.substack.com)

AISN #24: Kiss­inger Urges US-China Co­op­er­a­tion on AI, China’s New AI Law, US Ex­port Con­trols, In­ter­na­tional In­sti­tu­tions, and Open Source AI

Oct 18, 2023, 5:06 PM
14 points
0 comments6 min readLW link
(newsletter.safe.ai)

Back to the Past to the Future

PrometheusOct 18, 2023, 4:51 PM
5 points
0 comments1 min readLW link

How to Erad­i­cate Global Ex­treme Poverty [RA video with fundraiser!]

Oct 18, 2023, 3:51 PM
50 points
5 comments9 min readLW link
(youtu.be)

On In­ter­pretabil­ity’s Robustness

WCargoOct 18, 2023, 1:18 PM
11 points
0 comments4 min readLW link

At 87, Pearl is still able to change his mind

rotatingpaguroOct 18, 2023, 4:46 AM
148 points
15 comments5 min readLW link

(Non-de­cep­tive) Subop­ti­mal­ity Alignment

SodiumOct 18, 2023, 2:07 AM
5 points
1 comment9 min readLW link

mag­netic cryo-FTIR

bhauthOct 18, 2023, 1:59 AM
10 points
0 comments4 min readLW link
(www.bhauth.com)

Hints about where val­ues come from

Oct 18, 2023, 12:07 AM
24 points
13 comments10 min readLW link

Labs should be ex­plicit about why they are build­ing AGI

peterbarnettOct 17, 2023, 9:09 PM
214 points
18 comments1 min readLW link1 review

Eleuther re­leases Llemma: An Open Lan­guage Model For Mathematics

mako yassOct 17, 2023, 8:03 PM
22 points
0 comments1 min readLW link
(blog.eleuther.ai)

In­ves­ti­gat­ing the learn­ing co­effi­cient of mod­u­lar ad­di­tion: hackathon project

Oct 17, 2023, 7:51 PM
94 points
5 comments12 min readLW link

Wor­ld­work for Ethics

False NameOct 17, 2023, 6:55 PM
8 points
1 comment24 min readLW link

[Question] When build­ing an or­ga­ni­za­tion, there are lots of ways to pre­vent fi­nan­cial cor­rup­tion of per­son­nel. But what are the ways to pre­vent cor­rup­tion via so­cial sta­tus, poli­ti­cal power, etc.?

M. Y. ZuoOct 17, 2023, 6:51 PM
19 points
3 comments1 min readLW link